Nanostring Analysis of Lactate Correlation Geneset

  1. Filtering low count reads
    A. Determining Which Samples to Exclude
    B. Determining Which Genes to Exclude
  2. Compare COV of HKGs to lactate cor. genes
    A. Normalized nanostring data: comparing the COV for our 5 housekeeping genes to the COV of the genes that remained after excluding false positives
    B. RNAseq CPM data: comparing the COV for our 5 housekeeping genes to the COV of the genes that remained after excluding false positives
  3. Correlation of detected gene expression levels with 3-hour lactate levels
    A. Positively correlating genes
    B. Negatively correlating genes
  4. Correlation figures for probes that meet all of the threshold criteria
    A. Heatmap showing relative expression of the gene set detected by nanostring
    B. Plot showing showing lactate levels for each sample (red) with relative expression (grey) and mean relative expression (black) of the gene set detected by nanostring. error bars are the variance of the mean relative expression data.
    C. Mean expression between Adequate and Low performing livers for the 7 lactate correlation genes.

Summary

Of the 10 samples used, one was not usable (LV13) due to an error in sample prep. Another sample (LN11) was analyzed in triplicate to demonstrate the low level of variation between technical replicates. Data was normalized using internal positive controls and calculated in the nSolver software provided by nanostring. Housekeeping genes from the RNAseq data maintained a lower coefficient of variation then the 23 lactate correlation genes assessed. Of these 23 lactate correlation genes, only 12 yielded sufficient data above the false discovery threshold calculated using internal negative controls. Of these 12 genes, only 7 maintained a correlation of .8 with lactate levels as was observed in the RNAseq experiment.

Filtering low count reads

Excluding any probes that did not exceed the maximum count observed in the negative controls for that sample. This is the most stringent false discovery identification method suggested by nanostring

A) Determining Which Samples to Exclude

  • Excluding LV13 since none of the reads were greater than the negative control threshold

B. Determining Which Genes to Exclude

  • Excluding all genes with more than 2 false positives

Compare COV of HKGs to lactate cor. genes

Housekeeping genes should have generally lower coefficients of variation compared to the lactate correlation genes

A. Normalized nanostring data: comparing the COV for our 5 housekeeping genes to the COV of the genes that remained after excluding false positives

B. RNAseq CPM data: comparing the COV for our 5 housekeeping genes to the COV of the genes that remained after excluding false positives

Correlation of detected gene expression levels with 3-hour lactate levels

Excluding any gene that does not have a correlation of at least .8 between the nanostring data and 3-hour lactate levels

A. Positively correlating genes


Note: GSTA1 is only able to meet the cutoff when rounded. (r= 0.7977059)

B. Negatively correlating genes


There are 7 genes that meet the correlation cutoff of .8 that was previously used with the RNAseq data results. All of these genes positively correlated with lactate, with no negatively correlating genes having sufficient levels of correlation. Alternatively, setting the cutoff more leniently to .65 would still only yield a total of 10 genes which would include a single negatively correlating gene

Correlation figures for probes that meet all of the threshold criteria

A. Heatmap showing relative expression of the gene set detected by nanostring

B. Plot showing showing lactate levels for each sample (red) with relative expression (grey) and mean relative expression (black) of the gene set detected by nanostring. error bars are the variance of the mean relative expression data.

C. Mean expression between Adequate and Low performing livers for the 7 lactate correlation genes.